Document Dewarping via Curved Coordinate Net
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for capturing document images using digital cameras introduce image distortion, which is difficult to correct, especially when text lines are absent or the camera lens is not parallel to the document surface, limiting the effectiveness of existing models in dewarping and orientation correction.
Innovation Solution
A method and apparatus that utilize a stationary or non-stationary camera to capture images, involving image segmentation, distortion removal, and orientation correction, including the use of random sample consensus and outlier removal to identify document edges, and a dewarping process that applies a curved coordinate net to transform warped images into a flattened format, independent of text presence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a digital camera is used to capture document images, then portability and versatility are improved, but image distortion increases
Solution Approach 1:
The patent applies preliminary action by detecting document edges and determining the camera's angular orientation relative to the document surface before the dewarping process. This preliminary detection of edge coordinates and angle information enables subsequent correction of the image distortion caused by non-parallel camera positioning.
Solution Approach 2:
The patent employs parameter changes by using the detected edge coordinates and camera angle to calculate dewarping parameters that transform the distorted image coordinates into corrected coordinates. The system changes the spatial parameters of image pixels based on the determined angular orientation and edge positions to eliminate distortion.
2Measurement precision
If existing dewarping models are used, then text line detection is improved, but effectiveness deteriorates when text lines are absent
Solution Approach 1:
The patent extracts the document edge information from the image independently of text content. By detecting edges and determining camera angle relative to these edges, the system separates the dewarping function from text line detection, allowing effective operation even when text lines are absent or difficult to detect.
Solution Approach 2:
The patent creates a universal dewarping method that works for all document types regardless of text presence. The edge-based approach combined with angle detection provides a multi-functional solution that handles both text-rich and text-poor documents, as well as curved and flat surfaces, making the system universally applicable.
3Ease of operation
If the camera lens is not parallel to the document surface, then capturing flexibility is improved, but image distortion increases
Solution Approach 1:
The patent implements feedback by detecting the actual camera angle relative to the document surface from edge coordinates, using this information to calculate appropriate dewarping parameters. The system continuously adapts the correction based on the detected angular orientation, enabling effective dewarping regardless of the initial camera positioning flexibility.
Data Source
AI summary
An apparatus and method for processing a captured image and, more particularly, for processing a captured image comprising a document. In one embodiment, an apparatus comprising a camera to capture documents is described. In another embodiment, a method for processing a captured image that includes a document comprises the steps of distinguishing an imaged document from its background, adjusting the captured image to reduce distortions created from use of a camera and properly orienting the document is described.